Sensitivity of forecast rainfall verification to a radar adjustment technique

Abstract

A ground-based radar (GR) has to measure rain from close to the radar to large distances from it. Consequently, the scattering volume of the GR changes significantly. As an advantage, the scattering volume of a space-borne radar is of similar size at all locations, thus allowing the compensation of the decreasing spatial resolution of the GR with range (range-adjustment). Adjustment with range is here performed by means of data observed by the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) using a ∼10 dB per decade factor. For instance, about 8 dB are added to the measured reflectivity at 100 km, while 2 dB are subtracted at 10 km from the radar site. Thus, two different radar datasets, the range-adjusted data and the original ones, can be compared with forecast precipitation. In the framework of the EU VOLTAIRE project (Fifth Framework Programme), where observations from rain gauges, GR, TRMM PR and forecast precipitation were available for the island of Cyprus, such a kind of study was performed. The chosen comparison method was a contiguous rain area (CRA) analysis. Three pattern-matching criteria, involving mean square error, mean absolute error and correlation, have been used to match forecast and observed precipitation patterns. In this paper, we show that the results of the comparison in a selected case study are sensitive to the application of a range-adjustment technique. Observational analysis, obtained by merging rain gauge data with the adjusted GR data, seems to give more stable results when changing the pattern-matching criterion, and proposing it as the better field reconstruction in the comparison.